0nm
Here in the UK we already have 0nm fab. The chips we make are so tiny, nobody has seen them.
Japan's Rapidus broke ground on its IIM-1 plant in Hokkaido on Friday, kicking off a flurry of hiring as the foundry upstart races to bring its 2nm wafer fab online by 2025. During the ceremony, the company revealed they have already hired more than 200 people to bring the facility online on schedule, according to a Bloomberg …
... defect rate is equally important.
Today, chips are made of so many layers that require several operations: deposition, lithography, etching, doping, adding metal films, etc...
The total process therefore involves many steps of adding, modifying, heating, cleaning or subtracting layers in tiny layers during weeks.
There are many techniques involved, and to achieve your goal, you have to master all of them.
Some layers are deposited atom by atom (atomic layer deposition). Challenging...
The success rate of the whole chain is the multiplication of all successive success rates.
If each operation introduces 1% defect (success rate is 0.99), then after 500 operations, your overall success rate is less than 1%.
Which makes market worthy chips at the end of the chain very expensive.
So it's all about rigorous quality.
But that's not all. Because, whilst you were working on quality, so was the competition.
So it's more about how fast, compared to the competition, you reach an acceptable yield.
All in all, this is a race with very mature players.
Japan already has a number of market leaders in the industry (Tokyo Electron, etc), but still, this is a very ambitious goal.
How realistic and attainable, we shall see.
But if they fail, they won't be the first one and can always try again with more knowledge and experience.